The inventive concepts relate to devices, systems, methods, and/or non-transitory media for image processing. More particularly, various example embodiments relate to image processing devices and image processing systems capable of adjusting computational complexity of an interpolation filter, image interpolation methods, and image encoding methods.
In general, an image processing device improves the data transmission efficiency of an image by encoding or decoding the image and transceiving the same. An image, e.g., one frame, is divided into a plurality of blocks by an image encoding method. Predictive encoding of the image is performed by performing inter-prediction and/or intra-prediction on each block.
Inter-prediction refers to a method of compressing an image by removing temporal redundancy between frames, and a representative example of inter-prediction is motion estimation encoding. In motion estimation encoding, each block of a current frame is predicted by using at least one reference frame to be encoded. For example, a reference block that is most similar to a current block is found within a prescribed search range by using a prescribed evaluation function.
In order to more accurately predict blocks of frames, interpolation is performed on a search range of the reference picture, pixel values of sub-pixels smaller than integer-pixels are generated, and inter-prediction is performed on the generated pixel values of sub-pixels. Computational complexity needs to be increased to more accurately generate pixel values of sub-pixels, and the time and/or the size required for resources for an interpolation operation may increase.